Investors are increasingly looking to diversify their portfolios away from the heavily concentrated AI sector. The "Magnificent Seven" stocks, largely benefiting from AI advancements, now comprise nearly a third of the value of State Street's SPY fund, which tracks the S&P 500. When semiconductor companies are included, this figure rises to over 40%, raising concerns about overexposure to a single theme. Goldman Sachs' co-head and co-chief investment officer for multi-asset solutions, Alexandra Wilson-Elizondo, noted that institutional clients are actively seeking ways to mitigate their AI risk.
A new ETF, the Roundhill Heavy Assets and Low Obsolescence ETF (LOHA), launched in May 2026, exemplifies this "anti-AI" trend. Conceived by Josh Brown of Ritholtz Wealth Management, LOHA focuses on "heavy assets and low obsolescence" — companies with value in physical infrastructure, established distribution, and long-lived capital, rather than software. Its top holdings include Cummins (engines), AutoZone (auto parts), TFI International (freight), Lennox International (HVAC), and Newmont (gold mining). The fund aims to offer diversification at the business-model level for investors whose portfolios have become overly reliant on tech and AI, suggesting it as a 3% to 7% allocation.
While marketed as an "anti-AI" play, the irony is that some of these physical economy companies are indirectly benefiting from the AI buildout. For example, Cummins' Power Systems segment reported record Q2 sales of $2.3 billion, a 19% increase, due to selling backup diesel generators to AI data centers. This illustrates that even seemingly unlinked sectors can be influenced by the AI boom. Other diversification strategies include investing in dividend ETFs, utilities, international equities, or specific companies identified by UBS, such as McDonald's, PepsiCo, Charles Schwab, S&P Global, and SS&C Technologies, which offer defensive traits and lower valuations.
